1. Field
This invention relates to integrated circuit devices, and more particularly to a method and apparatus for digitally tuning the capacitance of integrated circuit components in integrated circuit devices.
2. Related Art
Capacitors are used extensively in electronic devices for storing an electric charge. As is well known, generally speaking, capacitors essentially comprise two conductive plates separated by an insulator. Capacitors are used in a plurality of electronic circuits including, but not limited to, filters, analog-to-digital converters, memory devices, various control applications, power amplifiers, tunable (also referred to as “adaptive” or “reconfigurable”) matching networks, etc.
One well-known problem to those skilled in the art of the design and manufacture of integrated circuits is the poor tolerance values associated with integrated circuit components, especially the tolerance values of passive circuit components. Due to process variations, device parameter spread, variations in critical parameters such as conductive layer sheet resistance values, film thickness, process uniformity and manufacturing equipment cleanliness, and other factors, integrated circuit passive electrical components often have tolerances that are approximately an order of magnitude worse than their analogous discrete external passive electrical components. Consequently, it has proven difficult and costly in the past to implement tuned networks or circuits using on-chip passive electrical components.
Post-fabrication trimming techniques can be used after manufacturing and testing an integrated circuit in order to physically alter the circuit using a variety of methods including “Zener-zapping”, laser trimming and fuse trimming. Disadvantageously, the prior art post-fabrication techniques produce only static solutions. Although the trimmed devices may perform adequately under nominal conditions, they may not perform adequately under all of the operating conditions of the integrated circuit. Therefore, methods for improving the tolerances of passive electrical devices in an integrated circuit are needed which do not require the use of post-fabrication trimming techniques. Further, an improved method and apparatus is needed which dynamically monitors and corrects the performance characteristics of integrated circuits under all operating conditions. The improved method and apparatus should monitor and correct the performance characteristics of tuned networks especially as these performance characteristics are adversely affected by poor tolerances of on-chip passive electrical devices, and by the variable operating conditions of the device.
FIG. 1 shows a prior art attempt at solving the problem of implementing an adaptively tuned circuit using on-chip passive electrical devices. As shown in FIG. 1, using an integrated switchable capacitor circuit 100, two terminals of an integrated tuned circuit (i.e., terminal A 101 and terminal B 103) can be selectively coupled to a bank of switchably connected capacitors (C1 through Cn). Each of the capacitors is selectively coupled between the terminals 101, 103 by closing an associated and respective coupling switch Sn. For example, capacitor C1 102 is coupled between the terminals 101, 103 by closing an associated switch S1 110. Similarly, capacitor C2 104 is coupled between the terminals 101, 103 by closing an associated switch S2 112. Finally, capacitor Cn 108 is coupled between the terminals 101, 103 by closing an associated switch S2 116. Because the individual capacitors are connected in a parallel configuration, the total capacitance between the terminals 101, 103 is equal to the sum of the individual capacitors that are switched into the circuit (assuming that the switches do not also introduce capacitance to the circuit). By electrically connecting the terminals 101, 103 to a tuned circuit that is on the same integrated circuit as the switchable capacitor circuit 100, the capacitors can be selectively switched in and out of the tuned circuit, thereby changing the capacitance between the terminals 101, 103 to a desired value. Thus, despite the potentially poor tolerance characteristics of the capacitors C1 through Cn, the tuned circuit can be adaptively adjusted to operate within desired parameters by simply changing the capacitance between terminals A 101 and B 103.
Disadvantageously, this prior art approach is undesirable when the tuned circuit operates at relatively high frequencies. For example, when the tuned circuit operates in the GHz range of operating frequencies, the bank of switches (e.g., 110, 112, 114, and 116) introduce significant loss into the tuned circuit and thereby degrade the circuit's performance characteristics. The prior art solution shown in FIG. 1 also disadvantageously increases both the amount of space (i.e., integrated circuit real estate) and the amount of power required to accommodate and operate the switches. Power requirements are increased due to the D.C. current required to operate the bank of switches.
As is well known, there is an ongoing demand in semiconductor device manufacturing to integrate many different functions on a single chip, e.g., manufacturing analog and digital circuitry on the same integrated circuit die. For example, recently there have been efforts to integrate the various mobile telephone handset (or cell phone) functions and circuits in a single integrated circuit device. Only a few short years ago, the integration of digital baseband, intermediate frequency (IF), and radio frequency (RF) circuitry on a single System-on-Chip (SoC) integrated circuit seemed improbable or nearly impossible owing to a number of factors such as incompatible process technologies, yield limitations, high testing costs, poor matching of passive components, and lack of on-chip passive components having adequate analog characteristics. However, a number of advancements have been made in circuit design, physical implementation of hardware components, process technologies, manufacturing and testing techniques. These advancements are making the integration of digital baseband, mixed-signal and RF circuitry into a single integrated circuit device more of a reality. One such advancement is described in an article entitled “Overcoming the RF Challenges of Multiband Mobile Handset Design”, by Mr. Rodd Novak, RF/Microwave Switches and Connectors, published Jul. 20, 2007, www.rfdesign.com. This article is incorporated by reference herein as if set forth in full.
As described in the Novak paper, the complexity of cellular telephones has increased rapidly, moving from dual-band, to tri-band, and more recently, quad-band. In addition, cellular phones need to be able to accommodate a variety of signals for peripheral radios, such as Bluetooth™, Wi-Fi, and GPS. This trend is expected to continue as other capabilities are added. As described in the Novak paper, handsets are now being developed that incorporate tri-band WCDMA and quad-band EDGE platforms. These architectures demand at least seven radios in a single handset. Complexity will continue to rise due to the increased popularity of peripheral radios and functions that also need access to the antenna. The increased complexity in mobile telephone handset design has greatly complicated the RF front-end by more than tripling the number of high-power signal paths. By its nature, a multiband handset must accommodate a plurality of RF signal paths that all operate on different bandwidths. Yet, all of the RF signal paths must share access to a single antenna. As described in the Novak paper, a very efficient solution is to route all of the competing RF signal paths to the antenna using a single single-pole, multi-throw, RF switch.
The assignee of the present application has developed and is presently marketing such RF switches, and exemplary RF switch designs are described in applications and patents owned by the assignee of the present application. For example, the following applications and patents describe RF switch designs that facilitate further integration of mobile handset circuitry: U.S. Pat. No. 6,804,502, issuing Oct. 12, 2004 to Burgener, et al., U.S. Pat. No. 7,123,898, issuing Oct. 17, 2006, also to Burgener, et al., (both patents entitled “Switch Circuit and Method of Switching Radio Frequency Signals”); pending U.S. application Ser. No. 11/582,206, filed Oct. 16, 2006, entitled “Switch Circuit and Method of Switching Radio Frequency Signals”; pending U.S. application Ser. No. 11/347,014, filed Feb. 3, 2006, and entitled “Symmetrically and Asymmetrically Stacked Transistor Grouping RF Switch”; U.S. Pat. No. 7,248,120, issuing Jul. 24, 2007 to Burgener, et al.; U.S. Pat. No. 7,088,971, issuing Aug. 8, 2006 to Burgener, et al.; pending U.S. application Ser. No. 11/501,125, filed Aug. 7, 2006, entitled “Integrated RF Front End with Stacked Transistor Switch”; and pending U.S. application Ser. No. 11/127,520, filed May 11, 2005, and entitled “Improved Switch Circuit and Method of Switching Radio Frequency Signals”. All of the above-noted pending patent applications and issued patents are incorporated by reference herein as if set forth in full.
While these advancements in RF switch design facilitate further integration of mobile handset circuitry, a significant problem is presented as a result of mismatched impedances present at the mobile handset antenna terminal. Due to the variable operational environment of the mobile handset causing the impedance at the antenna terminal to vary over a wide range, antenna impedance mismatch poses significant technical challenges for the mobile handset design engineer. The problems associated with antenna impedance mismatch are described in a paper entitled “Antenna Impedance Mismatch Measurement and Correction for Adaptive CDMA Transceivers”, authored by Qiao, et al., Published 12-17 Jun. 2005, by the IEEE in the 2005 Microwave Symposium Digest, 2005 IEEE MTT-S International, at Pages 4 et seq. (hereafter “the Qiao paper”), and incorporated by reference herein as if set forth in full.
As described therein, mobile handsets are used in a variety of configurations and positions, by users who manipulate the handset and, in particular, the antenna, in ways that are difficult to predict. While a nominal antenna provides an input impedance of 50 ohms, in actual usage the impedance at the antenna terminal can vary over a wide range, characterized by a voltage standing wave ratio (VSWR) of up to 10:1. (Qiao paper, see the Abstract). Consequently, it is a major design engineering challenge to maintain proper operation of the mobile handset over a wide range of antenna impedances.
For example, for the receiver, the non-optimal source impedance degrades noise figure, gain and dynamic range. For the power amplifier, the antenna impedance mismatch greatly impacts the efficiency, power gain, maximum output power and linearity. In the worst case, the high standing wave amplitude or possible oscillation caused by the mismatch in the circuit may damage the power amplifier. As described in the above-incorporated Qiao paper, in accordance with one prior art solution, an isolator, or Voltage Standing Wave Ratio (VSWR) protection circuitry, is inserted between the amplifier and the antenna in order to mitigate problems associated with the antenna impedance mismatch. Unfortunately, this solution is disadvantageous because it creates attenuation, and therefore decreases antenna efficiency. Other possible solutions include correcting the impedance mismatch using dynamic biasing of the power amplifier or using a tunable matching network. Adaptively correcting for environmental changes that cause antenna impedance variation (e.g. placing a finger on top of cellphone antenna) is an important motivation for the need for tunable components in handset RF front-ends. In addition, tunable components also allow the RF front-end to cover more and more frequency bands, without increasing the number of antennas in the cellular phone. One antenna needs to cover more frequency bands in the cellular phone. This has proven difficult to achieve in prior art mobile handsets. Using tunable matching networks, the performance of the amplifier can be preserved even under severe mismatch conditions. Several examples of tunable matching networks can be found in the prior art.
For example, exemplary tunable matching networks for use in mitigating problems associated with antenna impedance mismatch are described in a paper entitled “An Adaptive Impedance Tuning CMOS Circuit for ISM 2.4-GHz Band”, authored by Peter Sjöblom, Published in the IEEE Transactions on Circuits and Systems—I: Regular Papers, Vol. 52, No. 6, pp. 1115-1124, June 2005, (hereafter “the Sjöblom paper”). As described therein, adaptive (or reconfigurable) matching networks are used between the RF antenna and RF switch in order to continuously adapt to the changing antenna impedance. The adaptive matching networks described in the Sjöblom paper are implemented using a bulk CMOS process in a configuration using switched capacitor banks in conjunction with inductors. The capacitors and the inductors create a ladder network. On the antenna side, a voltage detector is followed by an analog-to-digital (A/D) converter. A controller system controls the adaptive matching network by switching the bank of capacitors through all possible combinations to arrive at a state yielding the best performance. FIGS. 2A and 2B show two exemplary prior art tunable matching networks (200 and 200′, respectively) made in accordance with the Sjöblom teachings. As shown in FIG. 2A, an exemplary tunable matching network 200 comprises a bank of three switched capacitors 202 coupled to an inductor 204 and a load 206. The load 206 typically comprises an RF antenna. To gain enough latitude to match a wide range of impedances, a single inductor will not suffice. An alternative prior art adaptive matching network 200′ is shown in FIG. 2B. The alternative network includes two inductors (204′ and 204″), and three capacitor banks (208, 210, and 212), arranged as shown in FIG. 2B, and coupled to the antenna 214. The inductors (204, 204′, and 204″) are typically located in “flipchip packaging” or low-temperature co-fired ceramic (LTCC) substrates.
Disadvantageously, the tunable networks described in the Sjöblom paper do not, and cannot be designed to provide sufficient power required by some wireless telecommunication applications. For example, the power handling capabilities of the tunable networks 200, 200′ are insufficient for mobile handsets designed for use in the well-known Global System for Mobile communications (GSM). In order to be able to be used in a GSM/WCDMA handset the tunable component needs to tolerate at least +35 dBm of power without generating harmonics more than −36 dBm (based on the GSM spec). Also the IMD3 (3rd order intermodulation distortion) for WCDMA needs to be sufficiently low (typ −105 dbm . . . −99 dbm). These are the same requirements that are imposed on handset antenna switches. The Sjoblom paper is designed for low power applications (typ +20 . . . +25 dBm). It uses a single FET and a capacitor, whereas the digitally tuned capacitor (hereafter, “DTC”) of the present teachings uses a stack of many FETs (typ 5-6) that improve the power handling capabilities of the DTC. Anything built on a bulk CMOS process cannot meet the higher power handling requirements. The UltraCmos process has the ability to allow use of stack transistors in the DTC thereby allowing the DTC to handle high power levels (similar to GSM/WCDMA antenna switches). Stacked transistors cannot be implemented using a bulk CMOS process due to problems associated with substrate coupling.
The above-referenced Qiao paper describes a tunable matching network 300 comprising silicon-on-sapphire (SOS) switches 302 coupled to shunt capacitors 304. An exemplary prior art tunable matching network 300 made in accordance with the Qiao teachings is shown in FIG. 3. As shown in FIG. 3, this tunable matching circuit comprises six transistors 302 which provide 64 (26) possible capacitor states. The best state is selected to meet any particular mismatch circumstance. The tunable matching network 300 is implemented on a PCB board using discrete components. The transistors 302 comprise 1000 μm*0.5 μm FETs arranged in parallel and combined by wire bonding. The ON resistance for the total switch is approximately 0.5 ohms, and the OFF capacitance is approximately 1.8 pF. While the switched capacitor approach taught by Qiao, et al., has promising aspects, an integrated circuit implementation using this approach would occupy significant integrated circuit real estate. For example, the die area estimate is approximately 1.2 mm2 per 0.5 ohm FET, which for a six bit switched capacitor exceeds 7.2 mm2 without the capacitors 204. A complete tunable matching network requires a total of four switched capacitor banks, leading to a total FET area of almost 30 mm2. In addition to the unwieldy die area required by the Qiao teachings, it is also difficult to accurately control the overall capacitance due to the tolerance differences in the discrete capacitors. The circuit also disadvantageously has inferior power handling capabilities, linearity and Q-factor values for some applications. In addition, in this prior art solution, degradation in performance is caused by parasitic inductance of discrete capacitors. It is advantageous to use integrated capacitors (as opposed to discrete capacitors) because the parasitic inductance and Quality-factor (Q) of an integrated solution is higher using an integrated circuit on a sapphire substrate than what is typically achievable using discrete SMD capacitors.
As described in both the above-referenced Qiao and Sjöblom papers, at higher frequencies using integrated circuit technology, much work has been done using Micro-Electromechanical Systems (MEMS) switches instead of CMOS switches and capacitors. MEMS switches, varactors and thin-film Barium Strontium Titanate (BST) tunable capacitors have been used in the design of tunable or switched matching networks. Disadvantageously, these approaches have disadvantages of cost, tuning range (also referred to as “tuning ratio”) (which generally corresponds with maximum available capacitance/minimum available capacitance), integration and linearity. For various reasons, these solutions fail to meet the power handling, tuning ratio, and linearity requirements imposed by many wireless telecommunication specifications. Even after years of research and development, several MEMS and BST manufacturing enterprises that were founded to pursue the tunable component opportunities have fallen short of the requirements and specifications set forth in various cellular telephone specifications. Consequently, mass produced tunable capacitors or inductors for GSM power levels (i.e., +35 dBm) and WCDMA linearity (IMD3 −105 dBm) simply do not exist. BST capacitors exhibit significant problems when operated at high temperatures where their Q-factor is significantly degraded.
For example, varactor diodes and bulk CMOS switched capacitors do not meet the power and linearity requirements of these cellular specifications. MEMS switched capacitor banks exist, but they do not seem to meet power and linearity requirements, they require separate high-voltage driver chip and hermetic packaging, and reliability is a problem in mobile handset applications. BST voltage tunable capacitors are based on ferroelectric materials. These prior art solutions have difficulty meeting power and linearity requirements. They also disadvantageously require an external high voltage (HV) integrated circuit in order to produce high bias voltages (e.g., 20-40V) and generally cannot be integrated with other control electronics. The BST voltage tunable capacitors also suffer from degraded performances due to hysteresis and temperature stability.
Therefore, a need exists for a method and apparatus for digitally tuning a capacitor in an integrated circuit device. A need exists for a method and apparatus that can overcome the disadvantages associated with the prior art solutions and that facilitates the integration of tunable capacitor networks on a single integrated circuit. The need exists for an apparatus that facilitates the full integration of a tunable matching network for use with other mobile handset circuits and functions. In addition, the need exists for an apparatus and method that can dynamically calibrate an integrated tuned capacitor network such as a tunable antenna matching network. The present teachings provide such a method and apparatus.
The details of the embodiments of the present disclosure are set forth in the accompanying drawings and the description below. Once the details of the disclosure are known, numerous additional innovations and changes will become obvious to those skilled in the art.